Schur complement solver

.gitignore

@@ -4,5 +4,12 @@ viz/
 # Rendered video outputs (only the GIFs in assets/ are committed, via Git LFS)
 *.mp4
 
+*.pyc
+
+build/
+build_tests/
+
+.cache/
+
 # Design notes are kept locally, not in the repository
 docs/design/

CMakeLists.txt

@@ -214,6 +214,7 @@ if(BUILD_TESTING)
     tests/motion_prior_perf_test.cpp
     tests/factor_manifold_facade_test.cpp
     tests/dense_linear_solver_test.cpp
+    tests/generalized_schur_solver_test.cpp
     tests/dense_cholesky_solver_test.cpp
     tests/dense_qr_solver_test.cpp
     tests/gauss_newton_test.cpp

README.md

@@ -82,7 +82,7 @@ See [`python/examples`](python/examples/README.md) to run them.
 | **Built-in factors** | Reprojection, PnP, between (SO(2)/SO(3)/SE(2)/SE(3)/Sim(2)/Sim(3)/SL(4)/vector), point-to-point, point-to-plane, symmetric point-to-plane, prior, constant-velocity/constant-acceleration motion priors (SO(2)/SO(3)/SE(2)/SE(3)) |
 | **Custom factors and states** | CuPy / NVIDIA Warp kernels in Python, or user-defined CUDA kernels via `SizedFactorBatch` / `SizedStateBatch` in C++; the same types work with every minimizer, including RANSAC — see [Custom factors and states](docs/sphinx/custom_factors_and_states.rst) |
 | **Numeric Jacobians** | Finite-difference Jacobians for any factor batch (manifold-aware, reuses each state's `Plus` retraction), selectable globally (`MinimizerOptions.jacobian_mode`) or per factor group (the `jacobian_mode_override` of `Problem.add_factor_batch`) — write a factor with only a residual and let cuNLS differentiate it; see [Numeric Jacobians](docs/sphinx/numeric_jacobians.rst) |
-| **Linear solver** | Block-sparse PCG (variable block-Jacobi preconditioner, default), NVIDIA cuDSS (optional, loaded via `dlopen()` at runtime — see [C++ Installation](docs/sphinx/installation.rst)), dense LDLT, dense Cholesky (cuSOLVER), dense QR (cuSOLVER) |
+| **Linear solver** | Block-sparse PCG (variable block-Jacobi preconditioner, default), [generalized Schur complement](docs/sphinx/generalized_schur.rst) (automatic state elimination), NVIDIA cuDSS (optional, loaded via `dlopen()` at runtime — see [C++ Installation](docs/sphinx/installation.rst)), dense LDLT, dense Cholesky (cuSOLVER), dense QR (cuSOLVER) |
 | **Safety checks** | Optional runtime validation (linear-solver diagnostics and more), off by default for low-latency solves — enable with `MinimizerOptions::disable_safety_checks = false` when debugging |
 | **Execution model** | Fully asynchronous via CUDA streams |
 

cunls/linear_solver/CMakeLists.txt

@@ -1,4 +1,5 @@
 add_library(cunls_linear_solver OBJECT
+  generalized_schur_solver.cu
   block_sparse_pcg_solver.cu
   block_tridiagonal_solver.cu
   dense_cholesky_solver.cu